Vigiliae Christianae 21 (1967) 150-163; North-Holland Publishing Co.




 [[Needs considerable editing, including adding the Greek, superscripts, etc. and more information on variants within G; now available on JSTOR, from which page scans have been inserted below.]]

According to the standard reference works and the most recent monographs dealing with the Epistle of Barnabas,\1/ the only direct witnesses for the Greek text of the Epistle are Tischendorf's Codex Sinaiticus (S and S\2), Bryennios' Codex Hierosolymitanus (H),\2/ and a family of MSS (G)\3/ descended from a common archetype in which Barn 5.7b-21.9 was welded onto Polycarp, Philippians 1.1-9.2 without any indication, with the resulting hybrid circulating under the name of Polycarp.  To these are added such indirect witnesses as the quotations made by Clement of Alexandria (Cl.A) and an old Latin version (L), now preserved [[151]] only in a 9th century copy, which may have originated as early as the year 200 (see below, n. 12). 

\1/E.g. J. Quasten, Patrology 1 (1950) p. 91; W. Schneemelcher, supplemented reedition of  Funk-Bihlmeyer, Die apostolischen Väter (1956) pp.  XXIII and LII; P. Prigent, L'epître de Barnabé I-XVI et ses sources  (1961) p. 10; and this writer's translation and commentary to Barnabas and the Didache (The Apostolic Fathers, ed.  R. M. Grant, vol. 3, 1965), pp. 17f.

\2/ Because this 11th century MS was discovered at Constantinople by Bryennios, in the library of the patriarchal monastery of Jerusalem, it received the designation "C" in some older treatments such as the standard critical edition of Barnabas by O. de Gebhardt and A. Harnack in Patrum apostolicorum opera 1.2 (1878\2).  Both Bryennios and A. Hilgenfeld, for whom Bryennios provided a collation, used the symbol "I" (τ εροσολυμιτικν χειργραφον).  The MS has subsequently been relocated in Jerusalem.  In the introduction to his editio princeps of the Didache in 1883, Bryennios included five pages of corrections to Hilgenfeld's 1877 edition  of  the Epistle, which first used the readings of H. These corrections have not been noted in subsequent editions of  Barnabas  in the Western world.  The readings for H given below have been confirmed on the basis of a microfilm  of  the  MS.

\3/ Since the oldest member of this family is the 11th[-13th] century MS Vaticanus gr. 859, which seems to most  closely represent the lost archetype (so Funk), some editors list only the evidence of "V" (e.g. Funk's 1901 edition and Funk-Bihlmeyer in 1956).  Actually, however, at least two [probably three] sub-groupings of MSS within the family seem to be present: G/1 (v o [f p]) and G/2, (b c n t).  Other MSS from the same mutilated archetype are known to exist (a s), but their affinities are undetermined.  A recollation of all the members of the family is needed to determine more precisely how they are related and whether, in fact, all except "V" can be ignored in the attempt to recover the readings of the archetype.  Gebhardt-Harnack describe MSS [b] c f n o p v. The present writer has examined thsee texts by microfilm.

Nevertheless, as long ago as 1925 there appeared among the papyri published in the series internationally known as PSI the transcription of a fragmentary page from a papyrus codex containing Barn 9.1-6.\4/ That it has gone relatively unnoticed until now is not surprising in view of the fact that virtually no systematic efforts have been made in the past few decades [as of 1967] to catalogue and present in usable (reference) form the overwhelming wealth of published Jewish and Christian papyri fragments.\4a/  Thus the various specialists seldom can feel confident about what has or has not been found.  I was informed of the existence of the Barnabas fragment by my colleague J. Smit Sibinga (Leyden; [then Amsterdam] ), who happened upon a photograph of the papyrus among other early Christian documents that had been displayed by the Biblioteca Medicea Laurenziana in 1964.\5/

\4/ Pubblicazioni delta Società Italianà per la ricerca dei Papiri greci e latini in Egitto: Papiri greci e latini VII (numbers 731-870), ed.  G. Vitelli (Florence, 1925), number 757 (pp. 40-43).

\4a/ See now [2008] J. van Haelst, Catalogue ...[etc.]. Also the surveys by Kurt Treu, and since his death (1990) carried on by Cornelia Roemer [details??]

\5/ This reproduction is included in Documenti dell' antichità cristiana esposti nella Biblioteca Medicea Laurenziana 6-30 Giugno 1964 (
Florence, 1964), tav.  XVIII (cf. also pp. 23f, no. 26).  Since I was unable to locate a copy of this publication in any of the American libraries at my disposal, I requested photographs of the papyrus (as well as microfilm of MS f in family G; see n. 3 above) from the Biblioteca Medicea Laurenziana, which were speedily and graciously supplied (see the Plate). 

According to its original editor, the papyrus dates from the late 4th or early 5th century, [but it may be even earlier].\5a/  This would make it almost equal in age to the codex Sinaiticus, and thus one of the oldest direct witnesses we have for the text of the Epistle.  The transcription in PSI VII, based on one made by R. Bianchi, is well done, and the textual notes accompanying it are entirely satisfactory for such a preliminary presentation of the fragment.  The present treatment carries no implied criticism of the work of the earlier editors, but aims at supplementing their observations and making this hitherto virtually unnoticed fragment available to a wider circle.  The transcription which follows is my own, based on the photographs and on the material in PSI VII; where the small sublinear "o" is found in the transcription, it indicates material that the PSI editors claim to have been able to decipher, with certainty or probability, but that is not legible in the photographs.  In the discussion and in the notes appended to the transcription, I have attempted to provide as much of  [[152 = plates (see below)]] [[153]] the relevant information as is necessary to appreciate the relation of the papyrus to the other texts of Barnabas.

\5a/ [In a letter to the author, dated 23 January 1968, Colin H. Roberts states "I suspect it has been dated too late; I should myself date it to the late third century. . . . It is not unlike some documentary hands of that period." Roberts had a tendency to date materials earlier than other paleographers, which needs to be noted in this connection.]

The fragment is 6.3 x 11 cm and is written in a flowing but irregular semi-unical hand with frequent ligatures.  Its provenance is unknown.  There seem to have been 21 lines per page, if the recto provides an adequate standard.  In general, there is no word division, but a rudimentary punctuation does appear in several places (cf. verso lines 4 [?], 7, 8, 9 [?], 10, 11, 14, 17, and 20; recto lines 5, 7, 10, 12, 13, 17, and 19), usually followed by an enlarged initial letter in the next word.  Accents are found over YMWN and PALIN (sic) in lines 16-17 recto. Supralinear "corrections" have been made, perhaps by the original transcriber, on line 6 verso (pwrrwqen) and line 6 recto (eipen).  The nomina sacra qeos and kurios (including oblique cases) appear in the unusual ab­breviated forms Q. (verso line 10; perhaps also recto line 10, unless the more normal QC is intended there) and K. (verso lines 8, 16, and 18; but not lines 4 [?] and 11).\6/ The only place where the horizontal supralinear line normally used to indicate such abbreviations is found is over the K. on verso line 8. On verso lines 9 and 15, ICTRAHL and OYPANE are spelled out in full, as, apparently, was PNEUMA on verso line 11.  There are several indications at recto lines 17-18 that the transcriber considered some sort of major break to have occurred (a new paragraph?) -- notice the marginal D (meaning "4" ?), the punctuation before palin, the horizontal stroke separating line 18 from line 17 at the left margin, and the extension of panta into the left margin (as in recto line 14, which may also be the start of a new quotation).  Because of the possibility of such reverse-indentation ["ekthesis"], it is difficult to be confident about how much text is missing to the left for any given line on verso; similarly, because the right margin of verso is so irregular, it is difficult to determine (by analogy) how much is missing to the right in each line of the recto.

\6/ See A. H. R. E. Paap, Nomina Sacra in the Greek Papyri of the First Five Cen­turies A.D. (Papyrologica Lugduno-Batava 8, 1959), who records no other instances of this type of abbreviation by suspension except his no. 283 (P.  Strassb. 254; 4-5 century), in which the suspension X occurs for Xristos.  Paap suggests that "the influence of the inscriptions" may be present here (p. 125 n. 5; see also pp. 100, 102, 126; the Banabas papyrus is no. 251 in Paap). [Colin Roberts' letter of 23 January 1968 confirms that "the nomina sacra are indeed peculiar; P. M. Fraser tells me that he doesn't know of any parallel in inscriptions, and I have as yet found none elsewhere."]

From the viewpoint of textual criticism, the papyrus fragment (P) is of great interest.  The textual witnesses for Barn 9.1-6 are MSS H [[154]] and S, which represent the same type of text throughout the Epistle; family G, with which the 7th century corrector of S (S\2) usually is allied; and the ancient Latin version (L), which shows some tendency to side with G (S\2) where the Latin is not hopelessly corrupt or idiosyncratic.  Unfortunately, no quotations from Clement of Alexandria (or other ancient sources) exist for this portion of the Epistle. [check TLG??]

The papyrus frequently supports the majority of witnesses against a unique reading in one of the witnesses; for example:

[[insert image of page "18"]]

against MS    H

9.2b     τοὺς αἰῶνας (see 18.2)] τὸν αἰῶνα P S G (L)

*9.3b-c     H reverses the order to 9.3c,b against P S G L

9.4b     47iriroi0eicrav] πεποίθασιν P S G

9.5a      ἡμῶν (bis)] ὑμῶν (bis) P S G L

against MS S

         *9.3d    icai li@ ii6vov is lacking in P H G L

                   9.4b    Ecy(pa@ev (S*) is corrected to ἐσόφιζεν in S\2 as in P H G

                   9.5b    7c6piTliTITat (=-TE)l irFptTli@OTIT& P H G (see also, where S* wrotc 7r6piTliTiTat and S2 corrected it to agree with H IEEPITII@07ITF,, while G has -"CYCTat or -0@aE(FOE)

                  *9.5c         &7c&piTliTITa is lacking in S, against P H G L

against family G\7/

*9.4a        h (ei G\vo) gar    @ y6.4d),?6Cx Kai @ P H S  / @rr fl, i@ C,') 

9.5a           6 Stj'65& P H S L

9.5a           6,KdvOag G\vopf, or 6,K(IvOav G\bcnt dK(ivOatg P H S


 \7/ Two additional passages occur, in which the evidence within family G is divided:    bpcbv     G\pfn*c\mg/t\c/] gcbv P S G\vo*bct*n\c/

9.3c   Oo(bagg G\bct cf. L clamoris ] Oocb@og P H S G@oPf-

against L\8/

9.1c          Kai (before 7CEPtTlI@OTJTP) is lacking, against P H S G  

9.3a          'rao'ca E:ig PaPT6ptov is lacking, against P H S G  

9.3c          TgKva is lacking, against P H S G

\8/ There are numerous instances of unique readings in L, but it seems futile to list them all since it is seldom clear whether the difference arises from corruption in the transmission of the Latin, from the techniques and/or carelessness of the translator, or from a faithfully represented Greek Vorlage (see the samples in the textual  notes).  Nor do the three readings listed here necessarily derive from the Greek Vorlage of L.

Occasionally, however, P contributes an unsupported reading of its own or follows a text which hitherto had relatively minor support; for example: [[155]]

readings unique to P\9/      ICUPIOU- (?) 06K @7E@KOucrav (?)] @n@Koucav H S G L 

9.3d     T6v k6yov] ?,6yov H S G

9.4b     6 nov7lp6g] novyip6q H S G

9.5a     6 Or.6q] ic6ptoq 6 0&6q H S G L

 \9/ To this list could be added the "corrections" made in the papyrus at nwp­pwoev for the normal spelling n6ppwOev (P- H S G), and at 9.4c [,,v (see L, dixit) for EIPTIKEV (P* H S G), as well as the interesting orthography in 9.2a ICTPAHA] IHA H G, ICA S, Israhel L. The spelling Istrahel or Isdrahel is not uncommon in certain (African) Old Latin MSS - see H.Rönsch, Itala und Vulgata (18752) p. 460, and C. H. Turner, Prolegomena to the Testimonia of St. Cyprian 11, J. Theol.  Studies 9 (1907/08) p. 78.  [ Ἰστραὴλ is also found in the Greek of 1 Enoch (Watchers) 10.1, PGM 4.1813, 5.108-116, 10.7, 36.310; Ἰσδραὴλ (and Ἰσδραηλιτῶν) is found 16 times in the TLG text of Didymus the Blind.] Indeed, Ἰστραὴλ is also found in certain "Western" Greek MSS of the NT (Matt. 19.28 W, Mark 12,29 DW, Luke 2.32 D) as well as several times in the book of Acts in codex B (see Blass-Debrunner-Funk § 39.5 [p. 22], and Moulton-Howard, Grammar ll,l [1919], p. 103).  The substitution of et=v for elpq­xev is especially interesting in view of Barn 10:1,3, where HS have El@ev but G has elpilwev (Barn. 6,19 is the only other occurance of elpilkev in the Epistle).

readings in P supported only by S

9.5c &KpopucyTiav] -crtia H, -crta G, corporis L

readings in P supported only by G

               *     7rEptgTcgsv @gC)v icai (?)] Kai is lacking in H S (L)

                    *9.5b   T6 GKkTlp6V 'C?N KapSia; (cf.  L nequitiam de praecordiis)) 'r@V OW),71POKaP5lav S   

*9.5c           Tc6tktv] ?,6,pc (sic!) ndkiv H S (cf.  L dicit autem iterum)

readings in P supported only by L\10/       Kai (Et @noiijau) = et (quaefaciet) ]Kai is lacking in H S G      6KO6,;=aures]icap8iaH

 \10/ See also the preceding note on the "correction" at 9.4c.

Furthermore, where P is preserved for readings on which the evidence of H S G L is split, it sides with L and its ally:

*9.2b   Kain(!,ktvT6 nv&6paKupiouirpo(pTITEOciPGL]Iacking in H S

*9.5a   ?,gy&t 5@ P H L I ?,gyzi S G


On the basis of this analysis of variation units in which the text of the papyrus actually is preserved, the following picture emerges (the complicated relationship to L as such is not included because of the special problems involved):


                        readings unique to P (4) or to PL (2)   =           6

                        readings unique to H                           =            5

                        readings unique to S (S2)                    =            4    (3)

                        readings unique to G                           =            4

                                                                                                19   total



                        agreement between P and HL alone      =          1

agreement between P and S alone         =          1

                        agreements between P and G (3) or GL (1)  =  4

                                                                                                6 total

 Thus of the 25 variation units under consideration (excluding "corrections" in P, orthography in proper names, and places where the G evidence is split), 19 are common to P and one or more of the other Greek witnesses.

                       In these 19 instances,

                                 P agrees with H         9 times

                                 P agrees with S\1    10 times

                                 P agrees with S\2    11 times

                                 P agrees with G        14 times

Furthermore, within this group of 19 variation units, 9 may be classified as "major" (quantitative difference, major transposition, use of different wording; these are marked with an asterisk above).  In this more limited context,

                      P     agrees with H (allied to another witness)             4 times

                     P     agrees with S (allied to another witness)             2 times

P       agrees with G                                                  8 times

                      P G are opposed to H S                                          4 times

                     P H S are opposed to G                                           1 time

It is thus obvious that, with reference to the Greek witnesses, P stands closest to the type of text known previously from family G (as the PSI editors recognized); this tendency is accentuated by the several agreements of P and Lagainst all or part of the other evidence. (The attempts to fill out the lacunae in the transcription of P below have been made with these considerations in view).  On the other end of the scale, Pis farthest from H, and from the H S alliance.  This provides an interesting contrast to the evidence drawn from the quotations of the Epistle by Clement of Alexandria (Cl.A).\11/ The extant texts of these quotations supply numerous otherwise unattested variants as well as 5 (or 6) unique agreements between Cl.A and L. In the 48 variation units in which Cl.A supports one or another of the Greek witnesses, however, there [[157]] is a definite tendency for Cl.A to agree with H (30 times; 25 "major"), whereas his text is farthest from S (21 times; 17 "major"), which often has unique readings caused by carelessness.  For the 43 variation units in which G is extant (that is, excluding Barn 1.1-5.7a), Cl.A and G agree 26 times (19 "major").

\11/ A collation and classification of this material is provided in the writer’s PhD thesis, The Epistle of Barnabas:  its Quotations and their Sources (microfilm, Harvard University, 1961; now available online in an unpolished draft ) pp. 32-40.  These 1961 listings have required slight adjustment and revision in view of the corrected readings from the microfilm of H (see above, n. 2), but the overall impression remains the same.

The main contribution of P to the textual criticism of the Epistle is the assistance it provides in focusing attention on what should already have been known or suspected on the basis of the other materials.

               ((1)  The type of text that developed into family G is at least as old as the H-S text, despite its present preservation in MSS dating from the 11th to the 17th centuries, all of which derive from a single, mutilated archetype.  The sporadic support given G by the 7th century corrector of Sinaiticus already pointed in this direction, and the not infrequent alliance of G-L against H-S should have made it clear even before the discovery of P. The mere age of Sinaiticus (c. 350 C.E.) does not in itself make the text of S more original or more valuable than the other witnesses. (In fact, the text of S is probably the most carelessly transcribed and least reliable in itself of any of the Greek witnesses.  On the whole, H seems more reliable as a witness to the H-S text type).  Both the H-S type of text, partly supported by Cl.A (whose support of G also is strong), and the G-P type, often allied with L (which also supports H-S against G on occasion), were available in the 4th, and probably in the 3rd century.

                      (2) The evidence of L, where it disagrees with H-S and G, should not be dismissed a priori as the product of mistranslation, emendation, or inner Latin corruption.  It is, certainly, difficult material to control with confidence, but the fact that both Cl.A and P preserve some significant agreements with L against the other witnesses suggests that a worthwhile residue of early variant Greek readings are yet to be distilled from the Latin.\12/

\12/ See the excellent edition by J. M. Heer, Die Versio Latina des Barnabasbriefs und ihr Verältnis zur altlateinischen Bibel (1908) pp. XL-LIX.  Heer considers a late 2nd century date for the translation to be possible, and argues that it is not later than the 3rd century (from Africa).  Other Latinists such as G. Bardy and C. Mohrmann also suggest a 3rd century date.

 (3)   Finally, the "unique" readings of P, added to those of Cl.A (and L), show that a much wider range of variation than is now attested by H-S and G was available in early MSS of the Epistle. [[158]]

[[Insert images of the papyrus]]

PSI      757     verso   (Barnabas 9.1-3c)


                            (1)  No trace of line 1 is preserved on the papyrus, but it has been supplied on the assumption that the verso, like the recto, had 21 lines.  At the start of 9.1 the Greek witnesses agree in reading yst-ydEp n&ktviccpic(bv eoricov.  All but the first word would fit comfortably on line 1. The slightly shorter equivalent to L fits even better and is printed above, although there is no reason to suppose that the first word of 9.1 necessarily began the page.

                            (2)  In line 2, the PSI editors print [7cgpigrF]@6V @ll(BV Kai T[@V Kap-' Siav] in agreement with most MSS of G (but Go"@hav'e bp6v).  The Kai is lacking in HS, while L has circumcidat aures praecordiae nostrae (= ? 'Cetq &KOCtg @]ICOV'Cfiq Kap8ia;), perhaps more correctly.  Chapters 9-10 of the Epistle are concerned with the idea of "circum­cised hearing" (i.e. exegetical gnosis), as is clear from (as pre­served in PL, see below), 9.3d (see below), and 10,12 in the Greek


PSI 757 recto (Barnabas 9.3d-6a) [[insert  scan of original page]]

                                               MSS (IrEptgTFIIFV T&g &KO&q @PLCOV Kai cbg KapSia;, Iva uuvi(bliev Ta,3Ta).  Since the text of P in rightly preserves the reference to "hearing", the conjecture that it also did so here gains plausi­bility.  From the photograph, nothing can be read with confidence after @li6)v (the final letter of which also is illegible); although what may be the lowest portion of an A appears about two spaces to the right of @g(bv.  On the basis of expected line length, there would be room for approximately 9 letters after hg(bv.  Thus Kai -c&.; &Ko(L; or Kai 'E@V Kap8iav are possible (in a relatively long line), assuming the correctness of the PSI editors in reading Kai, but the most preferable reading would be simply @g(bv Tdc; &Ko6,q - after which some early texts may have had the words Kai T@v Kap5iav (or cri@ icap8ia@, cf. 10,12), which contributed to the present textual problem here.  [[160]]

(3) 9.1b. Possibly line 3 was set slightly into the left margin (see recto), leaving ample room for the text of G @@7E:l 6 K6pioc, (HS lack 6; L has simply dixit [per prophetam).  For the quotation in lines 3-5 the PSI editors print Ei[@ ?]/[&Ko@v ? To]O KL)p@y- <sic> o6ic @TcilK[ / Ka]i... As they comment, the letters YKYP are certain;                                     (4)  so are the letters OYKOERIH, with the intervening IY, prob­able (on the basis of the photograph; the PSI editors note that Kt)pot, is impossible here).  HSG give a phrase from a variant form of the Greek Ps. 17,45a (18,44a) @ 2 Sam. 22,45b gig &KO@V TIOU @7ifIKoL)adv liou (L, auditu auris exaudivit me).  Most Greek OT MSS have 6n@KOU(74V 110t (@KOUGDV you in 2 Sam.), with such other variants as Ei,, @7taKO@v and @ir@Kouc7ev (-aag), but nothing like what is found in P. In 2 Sain. 22,42b (cf.  Ps. 17,42b), however, the words o6ic @7c@Kot)(Yev (or o6X bar-) (X6T(7ov do occur - "they cried ... to the Lord, and he did not hearken to them." P seems to represent a text in which the counter-idea is emphasized, that "they" (old
Israel) did not really "hearken" to what the Lord was saying (cf.  Isa 50,2; 65,12; 66,4).  Indeed, this idea not only makes sense in the context of Barn. 9.1-3, but is present in related OT and early Patristic texts which speak about Israel's inadequate "hearing/ears": cf. Ezek. 12,2 and Jer. 5,21, and especially the unique "Jeremiah" quotation in CI.A, Paed I,(9), 79:1 g(oq Tivog KEKP6.40tlat KCti 06K &iC7aKO6CTOVTUt; [- Hab. 1,2] i8O6 (ITCEPIT­JITJTA T(i j)TC! a@TCov [= Jer. 6,10] - "how long will I cry out and they will not hearken?  Behold, their ears are uncircumcised." Thus perhaps P had something like Pi; &Ko@v k6you icupiol) o6K gn@Kou­c;dv [toi) (lict) - "they did not hearken to me to the extent of hearing the word of the Lord" (see the reference to the "word" in 9.3d); or rig &.Ko@v @Tiot) Kupiou... - "they did not hear me with the hearing of the ear of the Lord" (i.e. circumcised hearing).  The word 6,KO@, and possibly also (bciov, is expected on the basis of and by analogy to most of the subsequent "quotations" of 9.1-3.

                             (5)    Quotations from Isa. 33,13 and something resembling

                             (6)            Jer. 4,4 (cL Deut. 10,16) are juxtaposed in lines 5-8.  The Kai of line 6 is found only in L; neither HSG nor the Greek MSS of Isaiah

                            (7) support it, although some OT witnesses prefix a icai to yv6)UOVTat i,-i line 7. In lines 7-8, the MSS of G read nEptrpqO@ar.Tat (vo*) or -0@cyr.(TOE (o-gpf bc@, while S* wrote 7tEpi-cpilTat (= The above reconstruction follows I-IS2 (cf.  L, cirmmcidite without the  [[161]] 

(8)    preceding et/Kai).  The reading &Kodg 6g6)v in line 8 probably preserves the original text of the Epistle; it is supported only by L (aures vestras), but is consistent with the theme of, 9.3d, and 10,12 (see above to line 2).  Nevertheless, HSG have c&.; Kup­5ia; OIAC)V here (cf., 10,12).

                      (9)        9.2a.  Perhaps line 9 also was set out into the left margin slightly (see recto) to accommodate the text attested by all witnesses Kai n&Xtv ?,@yet.  Otherwise the line would be slightly longer than the average lacuna suggests.  On the spelling tcrTpaTIX, see above, n. 9.

                        (10)      In line 10 there does not seem to be room for the text of HSGL

,c(18E: Xgyet K(6piog) 6 O(c6g) crou, but by omitting K- (as on recto, line 10), the expected length is achieved.  Of course, the writing might have been cramped here, or the line could have begun in the left margin, thus making room for the K-.

(11)      9.2b.  The formula of introduction in lines 10-11 is not found in HS, and is unique among the formulas of the Epistle.  Neverthe­less, both G and L support P here.  If it is a gloss, it is a very early

                        (12)      one.  The quotation in lines 12-14 seems to be based on passages

                        (13)      such as Ps. 33 (34),13 and Isa. 50,10a.  The only notable variant is in H, siq Tobg ai&vaq.

                        (15)      9.3a.  Both formula and quotation (Isa. 1,2a) in lines 14-16 are (17)      in exact agreement with HSGL.  In line 17, L does not have the concluding phrase TaD-ra F.!; paptf)ptov, found in HSG.  After liap,r@ptov, the papyrus seems to have a short horizontal stroke followed by the remnants of a large letter with what appears to be a rounded top, like a B. The PSI editors seem to have read it as K, for they print papT6ptov - <?> K[at.  If it is a B (and K looks less probable from the photograph), perhaps it represents the number 2, and on recto line 17 (margin) appears the number 4 - but what do they mean?

                              9.3b.              The order of quotations agrees with SGL (cf.  Isa. 28,14/ (18)      1,10, then Isa. 40,3) against H, where they are reversed.  Line 18 is almost illegible in the photograph, although the PSI editors print etKO6Ua-CE X6['YO]y

                        (20)      9.3c.  L has no equivalent for ),tyst or 'CgKVa in line 20, and (21)      reads vocem clamoris in line 21 (see G"',too6)cyTl;).  HS do not in­clude TI@ before (po)v@;.


                        (1)        9.3d (4a).  In line 1, H has ET&gcv and L circumcidite (as in 9. 1 c and 9.5ab bis) where SG read 7cEpttTEpFv.  LI(?) also has "corrected"  [[162]] 

(2    L* from nostras (@li6)v) to vestras (@pCov).  In lines 2-3, S uniquely reads dico6acopzv X6yov Kai li@ p6vov 7ciaT&@c7o)prv   (3)        @liE!;: none of the MSS have c6v before X6,lov, and G lacks the final word hiiet; (cf.  L).

                      (4)            9.4a (4b).  For HS &XX& Kai @ 7cE:ptTog@ (H, 7cepi Top@ !), G has h (or El in Vo) yetp ireptroli@.  This entire section is lacking in L. @(p' t itenoioaatv is supplied in agreement with G; S has @(p' @g

                      (5)        Tc&iroi0acrtv, and H has t(p' t tTcFnoi0etuav.

                                     9.4b (4c).  On the "correction" of E;TpTIKev (by the original tran-(6) scriber?), see above, n. 9. At the end of line 6 we would expect yev(v)ilO@vat with HSG (L has no verb), but that would make it unusually long by comparison with the other lines.  Perhaps the (7)        ending of the verb was abbreviated.  Similarly, 6ri 6,yyEXo@ in line 7 (8)        is longer than expected but is attested in all witnesses.  In line 8, anarthrous irov7lp6g is found in all the MSS, and S* uniquely reads ga(pa@E:v (corrected by S2).

                      (9)           9.5a.  The 69 (so HL) of line 9 is lacking in SG.  At the beginning (10)      of line 10 the PSI editors print [kg]/ygt 6 0- where all other wit­nesses to Barnabas have X@y&t K6pio@ 6 666q (cf.  Jer. 4,3, 7,3, etc.). In the photograph, however, the letters are not as unambiguously written as one could wish.  The mutilated initial letter could be A or T as well as r, then comes C with its horizontal midstroke extended in such a way that the next letter could be T or r as well as 1. The letters 0 and 0 are clear, but after 0 comes what appears to be a small 0 or C (rather than just a high dot) with a short hori­zontal stroke at the top right.  Thus a case could be made for read­ing [?,g]/y6t 6 Og or a corrupt text such as [k@] / YETO Oq or [6Tt Tdbc]/).t7<Ft> 6 &g.  H follows with @licov (see also line 13 below), not bli6)v as in SGL; then G has 8 5t not 656 (HSL); finally, for (11)      tvToX@v (line 11), L has noyam legem. In the phrase from Jer. 4,3b cited in lines 11- 12, the text of the papyrus agrees with HSG against the unique and possibly original. reading in L, vae illis qui seminant in spinis.  Where G has crn&ip7l EF,, HS read -Ecc (Greek OT MSS also are divided here); for @iri or

(12)      @7c' dx&,vOat;, some variation occurs within family G (see above, p.154), and many Greek OT MSS read tv for 6ic'.  Lines 12--13 give a phrase from Jer. 4:4a in which r(b Kupiq) is read by HGL against 

(13)      rq) Or.6) in S and most Greek OT MSS.  H reads @@Cov (see also line 10 above) with some Greek OT witnesses. [[163]]

9 Instead of the introductory formula of HSG, Kai 'ci L has the explanation (gloss?), hoc est, audite dominum vestrum.  In G and L, the subsequent quotation (cf.  Deut. 10,16) is introduced by KUi which would fill out line 13 nicely (although the fact that (14)      the initial letter of line 14 juts into the margin causes some hesita­tion - did the quotation syntactically begin there?).  In line 14, S reads TcFpixiiilTat (= -Ts) as in, and HS have c@v aK%TlpoKap- (15)      5i(xv with most Greek OT MSS, while L has nequitiam de praecordiis.

                                    In lines 15-17, Kai T6v cpdxilkov b@(bv o@ [i@ cyK%tlpuvF,!Tv. is essentially the text of G (which has c;i<),Tlp6vilTF,), whereas HS lack  (perhaps P did also) alid H appends ZTt (see LXX MSS in Deut.  (17)      10,16).  L lacks the entire clause.  The papyrus may have read CYK?,Tlpi)VETS in agreement with LXX MSS h and I* -there is no clear evidence for the i proposed by the PSI editors when they printed aK),ilpuvs[!]-cF..

9.5c.  The function of the marginal A is not obvious -is it a number ("4"; see also above to verso line 17), or an abbreviation for a correction (U, i8o@)?  None of the extant witnesses provide a text satisfactory for filling the lacuna at the end of line 17; all have a brief formula (rc&),tv G [Kai ndtktv G blq, ?,6LOP 7c6t?,tv HS, dicit autem iteritm L), followed by the words i5oi) ?,tyFL Kf)ptoq.  The PSI editors note that a grave accent seems to be placed over the last two letters of Tcakiv, and suggest that it may have been followed by & ),gyft (cf.  L).  Equally possible would be 7c6LXtv ?,ty&t K(i)ptoq).  There does not seem to be sufficient room in the line for ibo6.  In (18)     the quotation in lines 18-19 (cf.  Jer. 9.26b), 6,7cgpicliil-ca (HGL) is (19) lacking in S, but &Kpopuaciav is in agreement with S (and a quota­tion of the Jeremiah passage by Justin Martyr) against H (&KPO­Pu<yTiq.), G (dLKp6puaTrL), and L (corporis; cf. the reading of most   (20) Greek OT witnesses, cyapki).  Perhaps in the parallel phrase in line 20 we should also read Kap8iav (as in Justin's quotation of Jeremiah), although S has Kctp8ia; and HG have Kap8iq.Acf. L, cordis). (The PSI editors note that the A Of Kap8lt*l could also be M).

               (21)         9.6.  L lacks any equivalent for tpF.T@ and seems corrupt in what follows. 7rFpixtpvccat is supplied in agreement with G, while HS have Tc&pLTgTliriTut

Philadelphia 19104; University of Pennsylvania